Precision viticulture: monitoring and managing vineyard variation with remote sensing

Spatial variability within vineyard blocks can have a significant impact on the overall composition of harvested grapes and make forecasting difficult. Aerial imaging can be used to determine canopy size and condition of individual grapevines and thus ascertain a measure of variability in terms of the vine canopy. Image-derived, quantitative canopy descriptors have been shown to be correlated to yield, berry composition parameters and vine nutrient status. Consequently, the variation in a vineyard identified by an aerial image is useful in determining homogeneous management zones with different productivity characteristics. Specific management can be targeted depending upon the requirements of each homogeneous zone. In addition the spatial information can be used to increase the accuracy of sampling for ripeness and predicting final yield.While aerial vineyard monitoring has the potential to assist vineyard management during the growing season and at harvest, the technique can also assist research projects aiming to investigate vineyard variability or to determine the impact of treatments both of which alter vine productivity. In field experiments conducted since 1999, the National Wine and Grape Industry Centre has used airborne monitoring to assess treatment effects in previously uniformly-managed vineyards, which were applied to change vine productivity (vigour, yield). Early crop removal, post-harvest leaf removal or damage was easily distinguished by the remotely sensed images. We were also able to detect the impact of different irrigation treatments on canopy size with this technique. More recently, the most detailed study on the utility of remote sensing technologies in viticulture conducted to date has developed detailed canopy-characterising descriptors for use with ultra-high resolution (centimetres) remotely sensed imagery. Canopy descriptors quantifying foliage size and density were strongly correlated to anthocyanins, suggesting that this techniqcould also be used to segment a harvest based on fruit quality attributes.